BeagleBone USB FTDI device power independence

In my use case, it would be helpful if the FTDI USB to serial and JTAG
device could be powered from the mini-USB connector's 5V power, like
the USB hub is, and that the rest of the board could be configured to
only accept power from the external power supply connection.
Currently, the 3.3V for the FTDI device is generated in the TPS65217
and is the same power rail that supplies the ARM SoC. So, the FTDI
chip and the ARM SoC get powered together, either both are powered or
neither is powered.

I'd like to be able to connect my BeagleBone to my host PC, open my
serial terminal, and then power on the BeagleBone ARM SoC. In this
way, I can observe x-loader and u-boot running from a cold boot. With
the current configuration, either I miss the cold boot and I can press
the reset button to observe a warm reset, or I have to manually hold
the reset button down when connecting the USB cable and launching my
serial terminal (this feat requires 3 hands - 2 to type, 1 to hold
reset button). This 3 hand maneuver is required because the FTDI
device won't be recognized as a serial port by the host PC until it is
powered, and I usually can't launch a serial terminal if no serial
port exists.

My desire would mean that a 5V to 3.3V dc-dc converter would be needed
and its sole purpose in life would be to provide power to the FTDI
chip. Its addition would increase the cost of the BeagleBone and
possibly also the required board size / space, so that it could fit.
There would also need to be a reasonable way (removal of a 0-ohm
resistor?) of keeping the present power configuration (either USB or
external 5V can power entire BeagleBone) or using my desired power
configuration.

I understand the tradeoff being made. I don't disagree with the
current design for the common BeagleBone use case, but my use case is
one where observing and interacting with x-loader / u-boot from cold
boot is desired.

Thanks!

Thank you for your input. We will take your suggestion under consideration for future versions of the board. There are many trade offs to be made, cost being a big one. Adding cost and size of the board is not something we are looking to do. But, maybe we can figure out something where we can do this and figure how to handle the cost aspects.What you have suggested was a the beginning where we were, but was changed for several reasons, mainly powering the whole board from the USB port.

Gerald

If you remove FB6,7,8; stand R175,174,173,187,181 up on the FTDI side
pads, lift pin 6 on U13, it might be able to work on the current board
with a fixed 3 pin 5V in 3.3V out LDO regulator. But that's a hairy
rework. And I'm not familiar enough with the FTDI chip to know if the
VCCIO{A,B,C,D} (of which there's a typo for VCCIOC on the PDF
schematic) matter if they have no power but I'm thinking it won't.

A simple way to set it up for easy rework would be a 0-ohm supplying
VDD_3V3B to the FTDI chip. Then the rework could be simple as remove
0-ohm and wire up 3 lead LDO. That kind of rework would be doable and
shouldn't cost much to enable, assuming a 0603 0-ohm is cheap enough
and small enough and the layout isn't affected too badly with power
planes (haven't looked at the gerbers yet, sorry).

-Andrew

Could be. But, it would void the warranty, but you are certainly free to give it a try. We are ramping production and PCBs are ordered, so nothing will happen until after these next 18K worth are shipped out.

Gerald